Persistent effects of a tephra eruption (AD 1655) on treeline composition and structure, Mt Taranaki, New Zealand

Abstract

Tephra eruptions have significant long-lasting impacts on vegetation, and potentially explain extant vegetation patterns in volcanic landscapes. We quantified the effects of the AD 1655 Burrell Lapilli deposit, Mt Taranaki, on treeline vegetation. Where lapilli depth was 25–40 cm, a succession close to primary was initiated. Where lapilli depth was 5–25 cm, the canopy was opened, but some vegetation survived. Total tree basal area remains lower in affected vegetation (158 cf. 206 m² ha⁻¹), whereas total tree density is higher where a moderate disturbance stimulated regeneration (8433 cf. 6656 stems ha⁻¹). Griselinia littoralis and Podocarpus cunninghamii dominate treeline vegetation (c. 51 and 39 m² ha⁻¹, respectively) within the lapilli distribution. Where disturbance was moderate, a cohort of Libocedrus bidwillii was also initiated. Weinmannia racemosa is absent from treeline vegetation within the lapilli distribution, despite being a dominant component elsewhere (basal area 48 m² ha⁻¹). Compositional patterns result from interspecific differences in morphology and resilience, as well as light, substrate and temperature tolerance. Light-demanding, cold-tolerant taxa were able to take advantage of the newly created open sites, whereas shade-tolerant, less hardy species lost their competitive advantage at the treeline elevation. The successional trajectory of the treeline vegetation has been set back and altered, and there is no evidence of convergence.

Keywords:

• tephra
• Burrell Lapilli
• eruption
• disturbance
• vegetation succession
• New Zealand
• Libocedrus bidwillii
• Podocarpus cunninghamii
• Weinmannia racemosa
• Griselinia littoralis

Introduction

Volcanic eruptions represent infrequent and unpredictable disturbance events that have the potential to significantly affect vegetation. During eruptions, vegetation can be modified or destroyed by volcanic processes, and new geological substrates may be created for plants to re-establish upon (Dale et al. 2005a). Of all the vegetation disturbances that may result from volcanic eruptions, tephra fall is the most widespread in the landscape (Antos & Zobel 2005). Tephra is fragmental material explosively ejected into the air by an eruption and, depending on the diameter of fragments, can be categorised as ash (< 2 mm diameter), lapilli (2–64 mm diameter) or blocks and bombs (> 64 mm diameter). Eruption plumes containing tephra are readily influenced by wind direction, with the erupted material being transported and deposited downwind. Tephra deposits are generally deepest in close proximity to the volcanic vent, and become progressively thinner with increasing distance from the source (Carey 2005). The effect of tephra on vegetation is primarily related to the depth (thickness) of the ejected material (Eggler 1963; Vucetich & Pullar 1963; Hotes et al. 2004; Antos & Zobel 2005; Tsuyuzaki & Hase 2005; Gomez-Romero et al. 2006).

There is a close link between vegetation burial by tephra and succession pattern and process, with burials being retrogressive events that initiate either primary and/or secondary succession, depending on the severity of the eruption and the proximity to the source. Various studies have indicated that emergent, canopy and senescent vegetation are most susceptible to damage by tephra, whereas juvenile, robust or vigorous individuals have better chances of survival and are able to exploit forest gaps, thus facilitating forest succession (Kent et al. 2001). Vucetich & Pullar (1963) examined plant macrofossils in the volcanically active central North Island of New Zealand (Taupo Volcanic Zone), and attempted to determine critical tephra depths at which damage to vegetation would occur. They concluded that > 38 cm of tephra burial would result in complete destruction of trees, 20–38 cm in almost complete destruction, and 23–30 cm in partial destruction. On Mt St. Helens (Washington, USA), thinner 1–15 cm depositions of tephra were insufficient to kill canopy trees, but damage to the forest understory (i.e. shrubs, seedlings, groundcovers) could still be severe and influence the successional trajectory of the vegetation (Antos & Zobel 1985, 1986, 2005; Zobel & Antos 1997). Several confounding factors other than tephra depth also determine the effects that a tephra eruption will have on vegetation. These include differences in the stature, morphology and eco-physiology of the taxa involved, the availability of micro-sites for protection, tephra chemistry and temperature, and the occurrence of rain, snow and wind at the time of tephra deposition (Kent et al. 2001; Dale et al. 2005a). Antos & Zobel (1987) commented that data on vegetation changes caused by volcanic eruptions were sparse and mostly anecdotal, but the 1980 eruption of Mt St. Helens stimulated numerous succession studies (Dale et al. 2005b), as have recent eruptions in Japan (Tsuyuzaki 1989, 1991; Nakashizuka et al. 1993; Titus & Tsuyuzaki 2003), and elsewhere (Clarkson 1990, 1998; Whittaker et al. 1992; Oner & Oflas 1997).

In the North Island of New Zealand, numerous geologically young volcanoes (Rangitoto Island, White Island/Whakaari, Mts Tarawera, Ruapehu, Ngauruhoe, Tongariro and Mt Taranaki/Egmont) provide sites for the study of both primary and secondary succession initiated by volcanic disturbance. The analysis of plant macrofossils and pollen preserved in sediments across the North Island has often been used to determine broad-scale vegetation changes associated with volcanic disturbance (Vucetich & Pullar 1963; McGlone et al. 1988; Lees & Neal 1993; Clarkson et al. 1988, 1995; Wilmshurst et al. 1997; Horrocks & Ogden 1998; Giles et al. 1999), though given that the lifespan of many New Zealand trees is potentially > 500 years, present-day spatial differences in forest composition and structure can also be related directly back to volcanic events (Clarkson 1990).

In this study, we quantified present-day treeline vegetation across a relatively recent (AD 1655) tephra depth gradient on the New Zealand volcano Mt Taranaki. We asked whether, and to what extent, tephra depth is associated with present-day differences in vegetation composition, structure and dynamics. We also sought to determine the critical tephra depth thresholds for plant survival and recovery and whether species' differential responses to damage by tephra, and interactions between key species after the event, may have shaped the current vegetation pattern.

Study area and vegetation

Mt Taranaki/Egmont is a dormant andesitic strato-volcano, which dominates the landscape of the Taranaki Region, North Island, New Zealand (39°18′S, 174°4′E). Mt Taranaki rises 2518 m above sea level (a.s.l.), and alongside the progressively older volcanic remnants of Pouakai, Kaitake and Paritutu, defines the Quaternary volcanic lineament known as the Taranaki Volcanic Succession. The present-day cone of Mt Taranaki has been built up by eruptions in the last 10,000 years, and since c. AD 1500, Mt Taranaki has had at least 10 distinct eruption events (Neall 1980; Neall et al. 1986; Platz et al. 2012); although nearly all occurred prior to the European occupation of New Zealand, and no accounts are known from the indigenous Māori people (Lowe et al. 2002). The present study focused on the AD 1655 Burrell Lapilli eruption. Druce (1966) determined the year of the Burrell Lapilli eruption through dendrochronological dating of pre- and post-eruption trees, and was confident that the eruption occurred between AD 1654 and 1658, with the winter half of AD 1655 being the most probable date. The Burrell Lapilli deposit consists of up to 40 cm of loose white pumiceous blocks, lapilli and coarse ash (Topping 1972). Druce (1966) and Topping (1972) mapped the extent and depth of the Burrell Lapilli over Mt Taranaki, producing almost identical isopach maps. These maps indicate that tephra fallout occurred over an area of c. 150 km² to the east–southeast of the summit vent on account of the wind direction at the time, with a well-defined dispersal axis present from which the depth of lapilli gradually decreases from 40 to 0 cm. None of the other recent eruptions from Mt Taranaki have directly influenced the study site within the Burrell Lapilli distribution, other than the AD 1655 Burrell Ash and AD 1755 Tahurangi Ash, which are both distributed more or less evenly over the study area (Druce 1966). Soils on Mt Taranaki are predominantly recent/raw volcanic soils developed from andesitic tephra, and within the study area, Burrell Ash and Tahurangi Ash are the main parent materials (Tonkin 1970).

Mt Taranaki has a temperate maritime climate, with continually alternating calm and stormy weather conditions. Annual rainfall near the treeline elevation is c. 6000 mm, with an average summer temperature of 12–13 °C and an average winter temperature of 4–5 °C (Clarkson 1981). The slopes of Mt Taranaki are currently clothed with largely intact natural vegetation which has been described by Druce (1966, 1974, 1976a, b) and Clarkson (1977, 1981, 1986). The most obvious vegetation pattern relates to altitude, with an uninterrupted sequence from lowland forest through to alpine vegetation represented. The current study investigated treeline vegetation specifically. Here ‘treeline’ refers to the narrow belt of transitional vegetation between upper montane forest and subalpine shrubland, whereby the canopy is typically between 2 and 3 m in height and sporadically interrupted with taller emergent trees. On Mt Taranaki, this treeline occurs at c. 1050 m a.s.l. (Efford 2012) and the main trees represented are Podocarpus cunninghamii¹ and Griselinia littoralis, either with or without Libocedrus bidwillii, overtopping tree-sized shrubland species Brachyglottis elaeagnifolia, Pseudopanax colensoi and Raukaua simplex. Treeline composition here is comparable with many areas in the North Island (e.g. Pouakai Range, Mt Hauhungatahi, north side Mt Tongariro, south and southwest Ruahine Range) where Nothofagus is absent (Druce 1966). Below the treeline exists montane forest (600–1000 m a.s.l.) characterised by a c. 10 m high canopy of Weinmannia racemosa and lesser amounts of G. littoralis, intermixed with emergent P. cunninghamii and L. bidwillii. Above the treeline, vegetation stature gradually decreases to c. 2.5 m by 1100 m a.s.l., and crowns merge closely to produce a compact hedge-like canopy. Between 1100 and 1400 m a.s.l., tree species are replaced in the canopy by shorter shrub species including B. elaeagnifolia, P. colensoi, R. simplex and Hebe stricta var. egmontiana. Volcanic eruptions and associated processes on Mt Taranaki are thought to have destroyed c. 6700 ha of vegetation and damaged an equivalent area in the last c. 500 years (Druce 1966). Both Druce (1966) and Clarkson (1981, 1986, 1990) speculated that the contributions of W. racemosa, G. littoralis, P. cunninghamii and L. bidwillii may vary across the Burrell Lapilli depth gradient, although these differences have never been comprehensively quantified until now.

Methods

Data collection

Treeline vegetation was measured across the extent of the AD 1655 Burrell Lapilli deposit with 36 permanently marked 10 × 15 m quadrats (Fig. 1). Nested quadrat minimal-area checks had previously been used to determine an adequate quadrat size of 150 m² for surveying treeline vegetation in the study area (Clarkson 1981; J.T. Efford, unpubl. data). A stratified random sampling design was used to position the quadrats across three Burrell Lapilli depth categories (Druce 1966) at accessible treeline positions with 0–5, 5–25 and 25–40 cm lapilli depth. Twelve quadrats were located in each lapilli depth category. The 0–5 cm category included three quadrats from the treeline of the adjoining Pouakai Range, an area unaffected by the Burrell eruption. For ease of interpretation, the Burrell Lapilli depth categories of 0–5, 5–25 and 25–40 cm may henceforth be referred to as ‘unaffected’, ‘moderately affected’ and ‘severely affected’ areas, respectively; with the ‘eruption zone’ referring to the latter two categories only. For this study, ‘trees’ are defined as individuals > 2 cm diameter at ground height (dgh); tree ‘saplings’ are individuals < 2 cm dgh, but > 50 cm tall; and tree ‘seedlings’ are individuals 5–50 cm tall. In each 10 × 15 m quadrat, all trees were recorded by species and measured for dgh. The dgh of snags (standing dead trees; > 10 cm dgh) was also recorded. The physiognomy of the snags suggested they were either P. cunninghamii or L. bidwillii, however, because of their decayed state it was not always possible to distinguish the two, so these were grouped for analysis. In a randomly selected quarter of each quadrat (7.5 × 5 m), all tree and shrub saplings and seedlings were recorded by species and counted. Ground cover species in each quadrat were assigned a cover class based on the proportion of live foliage < 1 m, and all epiphyte species were recorded. A list of the vascular species present in each quadrat was also compiled.

Figure 1 Map of study site on Mt Taranaki (New Zealand), showing the location of the vegetation survey quadrats at the treeline position and the distribution of the AD 1655 Burrell Lapilli (after Druce 1966).

Data analysis

Tree species and snag basal area and density were calculated first for each quadrat, and then average (mean) values and standard deviations (SD) were obtained for each of the three lapilli depth categories. Diameter frequency distributions (histograms) were derived for key species in each lapilli depth category from pooled quadrat dgh measurements; histograms used seedlings as the smallest size class, followed by saplings, then a 2–10 cm dgh class, and progressive 10 cm dgh increments thereafter.

Non-metric multidimensional scaling (NMS), an indirect ordination technique, was performed in PC-ORD Version 6.0 (McCune & Mefford 2011) to graphically arrange the quadrats according to similarities in species composition and abundance alone (not constraining the axes to secondary variables). In the ordination analysis, the primary data matrix comprised the density values of 30 tree species found in 36 quadrats. No data transformations were considered necessary. The secondary matrix was composed of site variables specific to each quadrat, and included thickness of the Burrell Lapilli, elevation, slope and aspect. Site aspect azimuths were assigned to one of four classes, 1 for north (316–45°), 2 for east (46–135°), 3 for west (226–315°) and 4 for south (135–225°). Burrell Lapilli thickness was recorded both quantitatively (0–40 depending on lapilli thickness in cm according to Druce's 1966 isopach map) and categorically (using the three Burrell Lapilli depth categories previously described). Sorenson (Bray–Curtis) distance measure was used in the ordination and the analytical recommendations of McCune & Grace (2002) relating to stress value, stability criterion and number of iterations were followed. After ordination of the quadrats based on species composition and abundance, NMS independently assessed the relationship of the site variables to the species composition and abundance. A three-dimensional solution was selected and the ordination was rotated to obtain maximum correlation between axis 1 and Burrell Lapilli depth. Pearson correlation coefficients (r) calculated for each species and site variables, in relation to primary ordination axes, were used to assess variation in species composition and abundance across site variable gradients (Keith et al. 2010).

A permutational multivariate analysis of variance (perMANOVA) and post-hoc pair-wise comparison were used to test for significant differences in species composition and abundance between the three lapilli depth categories (Anderson 2001). The perMANOVA was conducted in PC-ORD using a Sorenson (Bray–Curtis) distance measure and the same primary data matrix used for the NMS ordination.

Results

NMS ordination

Some 57 vascular taxa were identified in the survey of treeline vegetation across the Burrell Lapilli deposit. All taxa were indigenous, constituting a mix of upper montane forest and subalpine shrubland species. Burrell Lapilli depth was the strongest influence on treeline vegetation composition (Fig. 2, Table 1). Axis 1 represented most of the variation in composition (R² = 0.42), and was strongly correlated to the Burrell Lapilli depth variable (r = 0.80). Topographic slope was also moderately correlated with axis 1 (r = 0.36). Axis 2 and axis 3 represented the remaining variation (R² = 0.36; 0.15), and were both correlated with elevation (r = −0.53; −0.45). No correlation between the ordination axes and quadrat aspect was apparent. The three Burrell Lapilli quadrat groups (severely affected, moderately affected, unaffected) were largely separated from one another in ordination space, with minimal overlap occurring between the groups. Several species displayed moderate (Pseudopanax colensoi, Alseuosmia macrophylla, Cyathea smithii) or significant (Coprosma grandifolia, G. littoralis, Coprosma tenuifolia, Melicytus lanceolatus) positive correlations with axis 1, suggesting that these species have become more abundant in treeline vegetation affected by the Burrell Lapilli (Fig. 3B). Conversely, significant negative correlation with axis 1 (Pseudowintera colorata, Myrsine salicina) suggests other species are suppressed from the Burrell eruption zone. A lack of correlation between axis 1 and several key species shown to differ in relation to lapilli depth results from nonlinear relationships.

Figure 2 Non-metric multidimensional scaling (NMS) ordination of treeline vegetation quadrats, Mt Taranaki. The plot was constructed using the densities of 30 tree species found in 36 quadrats. The ordination has been rotated to obtain maximum correlation between axis 1 and Burrell Lapilli depth. Axis 1 represents 42%, and axis 2 represents 36% of the variation in the data set. The third dimension (not illustrated) explained 15% of the variation. Unaffected = 0–5 cm lapilli; moderately affected = 5–25 cm lapilli; severely affected = 25–40 cm lapilli.

Figure 3 Photographs of treeline vegetation and Burrell Lapilli. A, Eastern sector of Mt Taranaki. B, Burrell Lapilli deposit (35 cm thick) near Dawson Falls. C, D, Treeline vegetation near The Plateau with emergent Podocarpus cunninghamii. E, F, Treeline vegetation near North Egmont with emergent Libocedrus bidwillii.

Table 1 Correlation (r) of species and site variables with NMS ordination axes.

	Axis 1	Axis 2	Axis 3
Species	r	r	r
Alseuosmia macrophylla	0.33	−0.13	0.04
Aristotelia serrata	0.13	−0.20	0.44
Brachyglottis elaeagnifolia	0.29	0.27	−0.55
Carmichaelia arborea	0.18	0.23	−0.43
Carpodetus serratus	0.11	−0.11	0.45
Coprosma grandifolia	0.73	−0.11	0.22
Coprosma pseudocuneata	0.11	0.12	−0.55
Coprosma dumosa	0.03	0.02	−0.49
Coprosma tenuifolia	0.53	0.01	0.24
Cyathea smithii	0.32	−0.05	0.40
Griselinia littoralis	0.71	0.35	−0.16
Libocedrus bidwillii	−0.24	−0.10	−0.40
Melicytus lanceolatus	0.50	0.10	0.27
Myrsine salicina	−0.30	0.63	0.14
Olearia arborescens	0.17	0.05	−0.36
Podocarpus cunninghamii	−0.16	0.53	−0.31
Pseudopanax colensoi	0.42	0.00	−0.41
Pseudowintera colorata	−0.80	−0.71	0.07
Raukaua simplex	0.11	0.09	−0.76
Schefflera digitata	0.23	−0.09	0.49
Weinmannia racemosa	−0.15	0.64	0.46
Site variables
Burrell Lapilli depth	0.80	−0.01	0.13
Elevation	0.05	−0.53	−0.45
Slope	0.36	−0.17	0.18
Aspect	−0.03	0.03	−0.01

Moderate (r = 0.3–0.5) and significant (r = > 0.5) correlations are given in bold. Species lacking correlation with any of the axes have been omitted from the table.

perMANOVA

A perMANOVA provided a statistical test of species composition and abundance across the three Burrell Lapilli depth categories and indicated a significant difference in vegetation (Table 2). Pair-wise comparisons distinguished that significant differences existed between unaffected and severely affected vegetation, as well as between moderately affected and severely affected vegetation.

Table 2 perMANOVA on treeline species composition and abundance across three Burrell Lapilli depth categories, Mt Taranaki. Highly significant P values (P < 0.001) are given in bold.

Source	d.f.	SS	MS	F	P
Burrell Lapilli	2	1.0335	0.51675	4.861	0.0002
Residual	33	3.5081	0.1063
Total	35	4.5416
Pair-wise comparison				t	P
Unaffected vs moderately affected				1.1985	0.2224
Unaffected vs severely affected				2.5177	0.0002
Moderately affected vs severely affected				2.6175	0.0002

Vegetation composition

Basal areas and densities of dominant tree species and snags are presented in Table 3. Across the three Burrell Lapilli depth categories, total basal area was greatest in the unaffected area (206 m² ha⁻¹), with progressively lower totals found across the moderately affected (180 m² ha⁻¹) and severely affected (158 m² ha⁻¹) areas. Total tree density was highest in the moderately affected area (8433 stems ha⁻¹), compared with the severely affected (6972 stems ha⁻¹) and unaffected (6656 stems ha⁻¹) areas, which both exhibited similar densities. The contributions of several potential treeline canopy or emergent species were found to vary considerably in relation to Burrell Lapilli thickness. Using basal area as a measure of species dominance, treeline vegetation unaffected by the Burrell Lapilli was characterised by G. littoralis (56 m² ha⁻¹) and W. racemosa (48 m² ha⁻¹), with lesser amounts of L. bidwillii (31 m² ha⁻¹) and P. cunninghamii (25 m² ha⁻¹). In moderately affected areas, G. littoralis (51 m² ha⁻¹), P. cunninghamii (43 m² ha⁻¹) and L. bidwillii (26 m² ha⁻¹) were most dominant, while W. racemosa (0.1 m² ha⁻¹) was essentially absent. In the severely affected areas, G. littoralis (51.1 m² ha⁻¹) and P. cunninghamii (35.1 m² ha⁻¹) were dominant, while W. racemosa and L. bidwillii were absent from the vegetation (Fig. 3). Snags contributed to the vegetation in all three lapilli depth categories. Basal area of snags was greatest in the severely affected category (27.3 m² ha⁻¹) and decreased with lapilli depth. Conversely, the density of snags was the lowest in the severely affected category (44 stems ha⁻¹) and did not correlate with lapilli depth; the moderately affected area had the highest snag density (122 stems ha⁻¹). Results show the high snag basal area in the severely affected category is a result of large diameter snags, in contrast to the moderately affected area, which had numerous smaller diameter snags.

Table 3 Basal area (± SD) and density (± SD) values for treeline species across three Burrell Lapilli depth categories, Mt Taranaki.

Burrell Lapilli depth	0–5 cm ‘Unaffected’		5–25 cm ‘Moderately affected’		25–40 cm ‘Severely affected’
Species	Basal area (m^2 ha^−1)	Density (stems ha^−1)	Basal area (m^2 ha^−1)	Density (stems ha^−1)	Basal area (m^2 ha^−1)	Density (stems ha^−1)
Podocarpus cunninghamii	25.3 ± 25.7	822 ± 567	42.7 ± 26.4	1300 ± 559	35.1 ± 22.7	728 ± 391
Griselinia littoralis	56.4 ± 49.7	228 ± 214	51.2 ± 29.3	533 ± 333	51.1 ± 40.1	683 ± 469
Libocedrus bidwillii	31.4 ± 58.1	83 ± 118	26.1 ± 29.0	156 ± 217	0	0
Weinmannia racemosa	47.8 ± 72.4	472 ± 750	0.1 ± 0.1	6 ± 19	0	0
Pseudowintera colorata	12.2 ± 9.3	3138 ± 2216	12.3 ± 8.5	3511 ± 1814	4.9 ± 4.1	1656 ± 1056
Raukaua simplex	5.4 ± 6.8	344 ± 436	8.6 ± 6.0	561 ± 573	4.2 ± 5.8	211 ± 234
Brachyglottis elaeagnifolia	0.2 ± 0.6	39 ± 78	4.9 ± 8.8	378 ± 794	2.3 ± 3.3	122 ± 192
Hebe stricta var. egmontiana	0.9 ± 1.4	128 ± 154	3.1 ± 4.5	317 ± 466	3.0 ± 6.7	167 ± 296
Pseudopanax colensoi	0.6 ± 1.3	50 ± 134	0.1 ± 0.1	17 ± 41	3.5 ± 8.4	122 ± 163
Coprosma grandifolia	0.1 ± 0.1	33 ± 60	0.2 ± 0.5	200 ± 346	2.7 ± 2.3	1256 ± 1308
Coprosma dumosa	1.2 ± 1.9	233 ± 282	2.1 ± 1.9	433 ± 300	0.8 ± 1.2	150 ± 131
Coprosma tenuifolia	0.6 ± 0.4	417 ± 321	1.2 ± 0.9	533 ± 390	2.7 ± 2.3	1061 ± 615
Myrsine divaricata	0.9 ± 1.1	167 ± 183	0.7 ± 0.7	122 ± 106	0.9 ± 0.7	189 ± 117
Myrsine salicina	4.6 ± 7.7	300 ± 564	0	0	0	0
Carpodetus serratus	0.7 ± 1.3	39 ± 66	1.0 ± 3.3	22 ± 43	0.6 ± 1.4	67 ± 211
Fuchsia excorticata	0	0	0	0	13.6 ± 40.4	83 ± 207
Aristotelia serrata	0	0	0	0	1.9 ± 3.8	83 ± 153
Cyathea smithii	0.2 ± 0.5	6 ± 19	0	0	2.4 ± 3.9	67 ± 106
Snags^a	17.8 ± 28.1	67 ± 74	24.5 ± 25.3	122 ± 94	27.3 ± 37.5	44 ± 70
Others^b	0.1 ± 0.1	89 ± 53	1.4 ± 0.7	222 ± 82	1.3 ± 0.5	283 ± 105
Total	206.4	6656	180.2	8433	158.4	6972

^a Standing dead Libocedrus and Podocarpus.

^b Alseuosmia macrophylla, Carmichaelia australis, Coprosma pseudocuneata, Dracophyllum longifolium, Melicytus lanceolatus, Olearia arborescens, Pittosporum tenuifolium, Rubus cissoides, Schefflera digitata.

A number of other less abundant tree and shrub species also displayed notable trends across the lapilli depth gradient. Fuchsia excorticata and Aristolelia serrata, for example, were only found within the severely affected area. Pseudopanax colensoi, C. grandifolia, C. tenuifolia, Carpodetus serratus and C. smithii were all more abundant in the severely affected area than elsewhere. The basal area of L. bidwillii and P. cunninghamii snags was greatest in the severely affected area. Brachyglottis eleagnifolia and H. stricta var. egmontiana were also more common in the eruption zone than the unaffected area. By contrast, M. salicina was only found in unaffected areas. Pseudowintera colorata, an important understory component across all lapilli depth categories, was least abundant in the severely affected area.

Diameter frequency distributions

The diameter frequency distributions (population structures) displayed by canopy species were variable across each of the lapilli depth categories (Fig. 4), and provide further insight into species response to the Burrell eruption. The area unaffected by the eruption serves as a reference point for populations that are unlikely to have had any recent influences from wide-scale disturbances. In the unaffected area, P. cunninghamii and W. racemosa both exhibited reverse-J type (continuous recruitment) population structures with high numbers of seedlings and progressively fewer stems in the larger diameter classes. Griselinia littoralis was represented by a range of diameters in the unaffected area (an all-aged population), but the majority of seedlings appeared to be failing to reach the subsequent sapling size class. Libocedrus bidwillii had a sporadic population structure in the unaffected area, which included a number of dead stems and an absence of seedlings. In the moderately affected area, P. cunninghamii and L. bidwillii both displayed cohort distributions, although only P. cunninghamii had representation in the seedling and sapling classes. Griselinia littoralis displayed a reverse-J distribution in the moderately affected area, with a higher level of seedling success apparent than in the unaffected area. Weinmannia racemosa was only represented by a small number of saplings and juvenile trees in the moderately affected area. In the severely affected area, W. racemosa and L. bidwillii were both absent, while P. cunninghamii and G. littoralis displayed cohort distributions, with several of the P. cunninghamii snags being > 1.0 m diameter.

Figure 4 Diameter frequency distributions of key treeline canopy species across three Burrell Lapilli depth categories, Mt Taranaki.

Discussion

Response of treeline vegetation to tephra damage

As a result of recent volcanic disturbances, the majority of the treeline around Mt Taranaki currently occurs at a lower elevation than on the adjoining Pouakai Range, with upper altitudinal limits of species often being higher on Pouakai (Clarkson 1990). The treeline position will slowly continue to advance up the slopes of Mt Taranaki, until it is again inhibited by another eruption, or species finally reach their potential upper altitudinal limits. Suppression of the treeline position has also been reported on Mt St Helens. Prior to the 1980 eruption there, the treeline was still advancing up the mountain in response to an eruption in AD 1800, and it was again suppressed considerably by the 1980 eruption (Swanson et al. 2005). More than 357 years after the Burrell eruption, structural and compositional differences were strongly evident in treeline vegetation across the Burrell Lapilli gradient on Mt Taranaki. This sector of the mountain also received ash deposition from the Newall and Tahurangi eruptions. The Newall event preceded the Burrell eruption by c. 150 years and resulted in c. 5 cm of ash being uniformly deposited over the study site (Druce 1961). This event would have increased soil nutrients but was unlikely to have had devastating or lasting effects on the vegetation, although minor canopy disturbance and seedling suppression could have decreased vegetation resilience to the effects of the Burrell Lapilli. The Tahurangi Ash was distributed across the upper slopes of the mountain c. 100 years after the Burrell event (Platz et al. 2012), where it has contributed to the formation of the current soil (Clarkson 1986). Although this ash may have had some impact on vegetation, given its concentric distribution, the effects would have probably been equivalent across the study area.

Treeline vegetation in the affected areas is yet to recover to the pre-eruption volumes indicated outside the Burrell eruption zone, with lapilli depth correlating with the numbers of trees destroyed by the eruption and the rate at which the vegetation has been able to recover. Large snags were more dominant in the severely affected area of the treeline. Natural senescence or possum damage cannot be fully ruled out as possible explanations for the trees deaths, however, their larger size compared with those in less severely affected areas, and lack of obvious root flanges, suggests these pre date, and were deleteriously affected by the Burrell eruption. Currently, total tree basal area is greatest outside the lapilli distribution, and decreases with lapilli depth. Total tree density was found to be greatest where the vegetation was moderately affected by the eruption (5–25 cm lapilli). Tree regeneration initiated by the eruption may, therefore, have been most prolific where lapilli damage opened the canopy to the extent that existing seedlings and saplings were released, or where substrate remained favourable for a colonisation event. Greater lapilli depths would have also opened the canopy, but few survivors would have remained and the substrate may have been unsuitable for new colonists for some time. Self-thinning is yet to return the total tree densities to pre-eruption stocking rates, with the successional trajectory of the vegetation having been set back. Disturbance events in forests, particularly those that open the canopy, are known to create a wave of regeneration resulting in an increase in stem densities (Denslow 1995). Canopy damage inflicted by 5–15 cm of tephra during the 1980 eruption of Mt St. Helens, for example, resulted in the density of tree seedlings increasing twofold, with much greater seedling establishment and survival than in the original undisturbed forest (Antos & Zobel 1986).

Differential species responses to damage

Vegetation would have responded variably across the Burrell Lapilli gradient following the eruption of Mt Taranaki, with mosaics created by a combination of the survival, death and arrival of species. Compositional patterns are thus likely to be a result of interspecific differences in, for example, morphology, resilience, demography and light and substrate requirements. Weinmannia racemosa, an abundant upper montane species over much of Mt Taranaki and at the treeline on Pouakai Range (Clarkson 1986), was shown to be absent from treeline vegetation severely affected by the eruption, and represented only by juveniles in the moderately affected areas. Previous research has determined that at lower elevations on Mt Taranaki, W. racemosa proliferated following eruption events due to its ability to resprout epicormically, and the copious amount of light wind-dispersed seed it produces (McGlone et al. 1988; Clarkson 1990). Weinmannia racemosa has also displayed a similar recovery following vegetation damage by the AD 1886 eruption of Mt Tarawera (Nicholls 1963; Clarkson & Clarkson 1983). Conversely, in treeline areas most affected by the Burrell eruption on Mt Taranaki, W. racemosa has not displayed this success. Species approach their upper-altitudinal limits at the treeline, and in this instance, the competitive advantage of W. racemosa to recover from disturbance appears to have been lost to more cold-tolerant species. On Mt Taranaki, P. cunninghamii, G. littoralis and L. bidwillii all display higher altitudinal limits than W. racemosa (Clarkson 1986), and thus have better exploited the eruption event at treeline. In parallel with Mt Taranaki, L. bidwillii was also able to dominate upper montane forest at the expense of W. racemosa following the AD 232 Taupo eruption (Hogg et al. 2011) on Mt Hauhungatahi, because L. bidwillii was more tolerant of the cooler conditions associated with exposure (Horrocks & Ogden 1998). Simulation of the post-Taupo vegetation dynamics on Mt Hauhungatahi using a forest landscape model also suggested a major shift in the montane forest composition, with initial L. bidwillii dominated stands being progressively replaced by W. racemosa following 400–700 years, when shade became a more limiting factor (Thrippleton et al. 2014). Myrsine salicina has also been excluded from treeline vegetation affected by the Burrell eruption, although it is quite common outside the distribution of lapilli. Myrsine salicina only appears in late-successional forest (Smale & Smale 2003), and is usually more prevalent on nutrient-depleted soils (B. D. Clarkson, University of Waikato, pers. obs.); the eruption has thus set back the successional trajectory of affected vegetation, with M. salicina yet to fully establish. Pseudowintera colorata was also found to be less successful in the severely affected areas, a pattern paralleled on Mt Tarawera where P. colorata is still rare in forest affected by the AD 1886 eruption (B. D. Clarkson, University of Waikato, pers. obs.).

Libocedrus bidwillii was suppressed from treeline vegetation, but in severely affected areas only. In moderately affected areas, it has been more successful with cohort populations being initiated by the eruption. Libocedrus bidwillii periodically establishes in even-aged stands following massive disturbance (Ogden et al. 2005), with seedlings being intolerant of shade and unable to regenerate below a closed canopy (Efford 2012). The absence of L. bidwillii from severely affected areas confirms that it was deleteriously affected by the Burrell eruption, but it is not clear why it has not responded with a typical wave of regeneration (Veblen & Stewart 1982; Boase 1988) such as that seen in the moderately affected area. Some possible explanations include that: 1. the deep lapilli substrate was unsuitable for L. bidwillii seedling establishment, 2. competitive exclusion by other light-demanding pioneer species occurred, 3. prevailing southerly winds in the severely affected southeastern side of the mountain inhibited L. bidwillii seed dispersal into the devastated areas, or 4. the timing of the eruption did not coincide with seed availability. The emergent physiognomy of L. bidwillii at the treeline would have resulted in it suffering the full impact of the lapilli shower, whereas species below the canopy may have been afforded some protection. Most emergent L. bidwillii at the time of the eruption never recovered from the stripping of foliage by lapilli, with some still remaining as decayed snags in the vegetation (Druce 1966; Efford 2012). Other studies have reported that L. bidwillii does not recover well from browsing damage by the introduced possum (Rogers 1997; Husheer 2006), further evidence that it may have struggled to recover from the mechanical damage inflicted by lapilli. Podocarpus cunninghamii also grows as an emergent in treeline vegetation, but is capable of more rapid foliar recovery than L. bidwillii following possum browse (Rogers 1997; Husheer 2006). Following the deposition of the Taupo tephra on Mt Hauhungatahi, L. bidwillii increased rapidly after the eruption in areas where tephra was < 9 cm thick, which is similar to the pattern on Mt Taranaki. At Mt Hauhungatahi, this was attributed to an increase in canopy gaps (created by the eruption), which permitted the light-demanding L. bidwillii to proliferate (Horrocks & Ogden 1998). Violent rainstorms which cause landslides can also be associated with large volcanic eruptions, and they too provide ideal sites for L. bidwillii colonisation (Boase 1988).

In treeline areas most severely affected by the Burrell eruption, P. cunninghamii and G. littoralis have become dominant. Perhaps not coincidently, both these species have bird-dispersed seed; as prevailing winds could have greatly inhibited seed transport into the area most severely affected by the eruption. Another possible explanation for the success of G. littoralis is related to its ability to colonise epiphytically. Elevated snags and debris would have been available immediately after trees were killed by the eruption, and the current growth form of many G. littoralis suggests that they have originated epiphytically. Clarkson (1981) also noted that G. littoralis had been establishing epiphytically on senescent L. bidwillii on Mt Taranaki, thereafter developing their own supporting stems to remain as canopy dominants. Griselinia littoralis is the sole coloniser of forest gaps in the montane zone at Waihaha and prefers epiphytic establishment (Smale & Smale 2003). Senescent G. littoralis commonly replaces itself in situ at Mt Pureora, whereby seedlings establish epiphytically in crowns of parent trees, thus prolonging dominance (Smale & Kimberley 1993). A reciprocal regeneration cycle between G. littoralis and P. cunninghamii similar to that discussed by Lusk & Ogden (1992) could also be occurring on Mt Taranaki, with G. littoralis capturing gaps formed by P. cunninghamii windfall, the latter, in turn, regenerating beneath the thinning crowns of older G. littoralis. On Mt Tarawera vegetation recovery has been slow, with P. cunninghamii only just beginning to colonise c. 110 years after the AD 1886 eruption, with facilitation by other species including G. littoralis thought to be important factors aiding its establishment (Clarkson & Clarkson 1995). Estimated ages for P. cunninghamii within the study area suggest it did not colonise immediately after the eruption, instead requiring a period of c. 150 years to establish (Efford 2012); probably time for suitable substrate to form as some lapilli was removed by erosion and soil developed.

Vegetation succession

Considering both the critical tephra thicknesses reported by Vucetich & Pullar (1963), Tsuyuzaki (1989) and Dale et al. (2005b), and the results of the current research, it is likely that where the effects of the Burrell Lapilli were most severe (i.e. 25–40 cm lapilli), the large majority of trees and shrubs were killed, and a succession close to primary was initiated. A suite of species found to be more common in treeline vegetation within the Burrell eruption zone (Brachyglottis elaeagnifolia, H. stricta var. egmontiana, Pseudopanaxcolensoi, Raukaua simplex, Coprosma spp., Carpodetus serratus, Fuchsia excorticata, Aristolelia serrata) are mostly subalpine shrubland vegetation constituents above treeline (Clarkson 1986) or seral light-demanding taxa. Pollen analysis of sediments from the eastern flank of Mt Taranaki suggest that seral shrub species, particularly Coriaria arborea var. arborea, increased in abundance following the Burrell eruption (Lees & Neall 1993). Shrubland vegetation is essentially a successional precursor to upper montane/treeline forest on the mountain (Boase 1988), and the Burrell eruption disturbance has thus acted to suppress the elevation of the treeline to an extent, by setting back and retarding the rates of species succession. Where the Burrell Lapilli depth approached 40 cm, compositional and structural differences are still markedly apparent in the treeline vegetation. Given the time that has passed without convergence it is likely these differences will persist. Similarly, during the AD 232 Taupo eruption (Hogg et al. 2011), < 10 cm of tephra was thought to have damaged vegetation via mechanical stripping of branches and foliage, with the crowns of canopy and emergent trees being most affected, and mechanical damage exacerbated further with chemical damage by the acidic tephra. Within 200 years of the eruption, revegetation of areas totally overwhelmed was completed, and post-eruption forests were similar to pre-eruption forests (Wilmshurst & McGlone 1996). However, closer to the source of the eruption, on the Hauhungaroa Range, tephra depth approached 50 cm and recovery of vegetation was slower, taking up to 450 years. Unlike the more distal sites, vegetation composition was changed more permanently, with, for example, P. cunninghamii becoming abundant and L. bidwillii becoming rare after the Taupo eruption (Clarkson et al. 1985). Seedling composition on the forest floor also changed following the addition of tephra on Mt St. Helens; a result of a partially intact tree canopy with a drastically altered substrate. Often following a disturbance, species not previously present in the vegetation will invade, but on Mt St. Helens, despite nearly 100% plant death in some locations, few species absent from the pre-disturbance forest established. Instead, there was a compositional divergence of the existing species, with some recovering rapidly while others failed to increase. This resulted in a complex mosaic of survival over the landscape (Dale et al. 2005a). In the decades after the event, this compositional shift has not been towards pre-disturbance conditions (Antos & Zobel 1986, 2005). The treeline vegetation on Mt St. Helens is now an unusual mix of conifers and hardwoods, which differs considerably from adjoining ranges which were not affected by such recent volcanic disturbance (Swanson et al. 2005).

The introduction of browsing mammals (goats, Capra hircus; brush-tailed possums, Trichosurus vulpecula) to Mt Taranaki has also had an effect on the successional trajectory of the vegetation in the study area, reducing the vigour and abundance of palatable species while promoting the proliferation of less palatable types (Clarkson 1990; Husheer 2006). Goats were first recorded in the park in 1910, with numbers peaking during the 1960s (New Zealand Forest Service 1975; Forsyth et al. 2003) when damage to the vegetation became so severe that ‘dead areas’ as great as 5 ha developed in lower altitude forest (Clarkson 1990). In the upper altitude forest, effects were not as great, however, compositional changes were still apparent, with depletion of G. littoralis, F. excorticata, P. colensoi and R. simplex, whereas the less palatable shrub P. colorata, as well as grasses and ferns became more abundant. Plant species capable of epiphytic regeneration such as W. racemosa and G. littoralis may have partially avoided goat browse, and this also has potential to change the vegetation profile; historically these species may have had much higher success establishing on the forest floor. Many plant species (e.g. W. racemosa, P. cunninghamii, L. bidwillii, F. excorticata, P. colensoi, M. robusta and R. simplex) present in the study area are also vulnerable to possum browse and, as elsewhere in New Zealand (Smale & Smale 2003), possums represent a contemporary influence on vegetation succession. Locally, heavy possum browse of W. racemosa occurred where Pseudopanax spp., F. excorticata and A. serrata were abundant and provided food variety (Clarkson 1990). This may have further confounded W. racemosa recovery in the areas severely affected by the Burrell eruption, where these palatable species were more dominant. Significant reductions in browsing mammal numbers recently occurred, primarily a result of intensive pest control beginning in the 1970s, including recent aerial broadcasting of 1080, and this led to dramatic recovery of vegetation in some areas (Clarkson 1990, 2011; Husheer 2006). However, despite pest control W. racemosa continued to decrease in tree density (Husheer 2006). Weinmannia racemosa could potentially regain dominance at the treeline within the eruption zone, although given the considerable time that has elapsed this is improbable. Herbivore pressure, not present prior to human arrival, coupled with W. racemosa having lower climatic tolerances compared to other species (e.g. L. bidwillii, P. cunninghamii), likely limits its recovery at the treeline within the Burrell eruption zone.

Our research has quantified differences in treeline composition and structure directly resulting from the AD 1655 Burrell Lapilli distribution. Emergent species appear to have been the most heavily affected by the eruption, and where tephra depth approached 40 cm, it is likely that almost all vegetation was killed, permitting seral and light demanding taxa to proliferate. Although the treeline position is expected to advance up the mountain until species reach their upper altitudinal limits, the successional trajectory of vegetation has been set back and altered by the eruption to the extent that differences may persist indefinitely and convergence is unlikely.

Acknowledgements

We thank the Department of Conservation for Permission to undertake this study on DOC-administered land. We also thank the George Mason Charitable Trust and the Environmental Research Institute, University of Waikato, for providing funding for this project, and two anonymous reviewers for providing constructive comments on an earlier draft of the manuscript. This research is dedicated to the memory of Anthony (Tony) Peter Druce, for his pioneering research on the vegetation of Mt Taranaki.

Notes

1. Nomenclature follows Ngā Tipu o Aotearoa (http://nzflora.landcareresearch.co.nz/).